Case Study: Bayt.com

The Short Version

• The largest job platform in the Middle East, sending 220 to 260 million emails a month through a mail server with no metrics, no bounce handling and no priorities. Email traffic was falling and nobody could say why.
• I stopped the bleeding at Microsoft, built the first bounce management in the company's history, then moved the whole operation from Postfix to PowerMTA: four message streams, quiet hours, a send-rate cap.
• 90%+ inbox placement, open rate up 140%, email-driven traffic up 150% year over year. On fewer emails.

What They Do

Bayt.com is the largest job platform in the Middle East: over 30 million users across MENA, job seekers on one side, employers on the other. At peak, two to three sends a day to the whole base. 220 to 260 million emails a month.

My Role

Brought in to fix a declining email channel. I rebuilt the email infrastructure end to end, then took over the full job seeker product experience.

The Challenge

Postfix. Zero visibility. No metrics, no bounce processing, no send-rate control, no way to tell a job recommendation from an ad. Bounces were ignored, so dead addresses got mail forever. The only signal anyone had was email referral traffic in Google Analytics, dropping month over month, steepest from Hotmail.

• No bounce management: hundreds of thousands of dead addresses hitting sender reputation
• One queue: transactional, job recommendations, marketing and third-party ads at equal priority, no rate limits, no quiet hours
• Microsoft was suppressing them: reputation damage under SRD vote policy changes, throttling the whole Hotmail, Outlook and Live family
• No double opt-in, and many teams sending to one database with no standards

A 220 million email a month operation that had never been blocked. I still don't know what saved them. It wasn't good practice.

What We Did

Stop the bleeding

Postfix couldn't give me domain-level data, so I couldn't size the Microsoft problem. GA could show me where the drop was. I hard-coded a stop on every non-critical send to Microsoft-hosted addresses: transactional and job recommendations kept going, marketing and ads paused. Not elegant. It stopped the damage while we built the real fix.

Bounce management, by hand

I classified about 500 bounce codes into soft and hard myself and set a rule: five softs equal a hard, suppress. The dev team wasn't thrilled. It was the first bounce management Bayt had ever had. A syntax check at registration and Kickbox before send kept new bad addresses out.

Postfix to PowerMTA

About $25,000 a year, sold to the infrastructure team, the dev team and people who'd never heard of an MTA on what staying was costing them in traffic and reputation. What it bought:

• Four message streams by priority: transactional, job recommendations, strategic marketing, third-party ads. Each with its own rules, limits and reputation. A job recommendation could no longer sit behind an ad campaign.
• Real bounce handling. Every ISP has its own taxonomy. PowerMTA knows thousands of codes. My spreadsheet knew 500.
• Quiet hours and a send-rate cap of roughly 1.2 million an hour instead of bursting the whole queue.

We sent far fewer emails. Every metric went up. "Upload a photo to your CV" was never worth sending; "here are your recommended jobs" was, and now it landed.

Standards across the company

My manager was the General Manager. That reporting line is the only reason I could walk into ad sales and tell them their campaigns were paused. Email usually sits too low on the org chart for this. I used the authority, then ran workshops for every team that touched email: frequency, reputation, what deliverability means in practice.

Lifecycle and the MENA problem

Rebuilt the job seeker lifecycle (onboarding, retention, re-engagement) from scratch and killed the campaigns that weren't earning their place. Then the region: Dubai email-native and app-first, Egypt SMS-first with low confirmation rates. Arabic, French and English. RTL and LTR in the same templates. A dev team in Jordan, so six weeks a year of Ramadan when heavy refactoring doesn't happen. You plan around it.

Results

• 90%+ inbox placement, no leased infrastructure or certifications
• 140% increase in open rate
• 150% increase in email-driven site traffic year over year
• Microsoft deliverability fully restored, and the first working bounce management in Bayt's history
• Promoted to lead the full job seeker product experience

What This Project Taught Me

Scale doesn't change the principles. It changes the blast radius. Every decision touched 220 million emails, and getting Jordan, infrastructure, ad sales, content and a GM to agree on sending standards was a different job from writing good flows. Harder, mostly.

Stack

PowerMTA · Postfix (inherited) · ReturnPath Inbox Monitor · Litmus · Kickbox · Google Analytics · Custom bounce classification system · Multi-stream send architecture · RTL email coding